GPs’ motivation for teaching medical students in a rural area—development of the Motivation for Medical Education Questionnaire (MoME-Q)

Design and GP recruitment

We conducted a study in which all general practitioners in the district of Siegen-Wittgenstein (n = 158) were invited to take part. Contact details are freely available on the website of the Association of Statutory Health Insurance Physicians in the region of Westphalia (https://www.kvwl.de/earzt/). GPs were asked regarding their motivation to participate in medical education of students as well as their work satisfaction and burnout risk. In addition, they were asked in detail regarding their future prospects of working as a GP. In this paper, we focus on the GPs motivation to participate in medical education of students.

This survey (HaMEdSi: Hausärzte (GPs) for Medical EDucation in Siegen-Wittgenstein) was performed in general practices in the area of Siegen-Wittgenstein in Germany between October 2017 and January 2018. GPs were sent a written invitation with a detailed study description, informed consent and the study questionnaire. After 4 weeks, all GPs who did not respond received a telephone reminder by a member of the study team. An invitation to participate was also sent by email to all members of the local doctor’s association, in which most of the GPs hold a membership. Furthermore, an informative meeting on the medical education perspective at the University of Siegen was held to which all GPs were invited and n = 45 took part. In this meeting, GPs were also reminded and invited to take part in the study.

The study was performed in accordance with the Declaration of Helsinki and approved by the research ethics committee of the University of Marburg (Az.: Studie 127/17).

Development of the Motivation for Medical Education Questionnaire (MoME-Q)

As there is no appropriate tool available to assess GPs motivation to take part in medical education of students, we developed a questionnaire based on the existing literature (Thomson et al., 2014; Ingham et al., 2015) as already mentioned above. Further items were developed and consented by expert panel meetings involving GPs as well as medical specialists experienced in the training of medical students. Afterwards a small pilot study (n = 6) was conducted among GPs experienced in the training of medical students (n = 3) as well as among GPs with less experience in teaching students (n = 3). All participants were invited to give a detailed written feedback and were also interviewed for further feedback on the questionnaire. The initial version of the questionnaire was slightly adjusted after the pilot study and as a result the Motivation for Medical Education Questionnaire (MoME-Q) developed. The MoME-Q is a 24item questionnaire with a four-point Likert scale with verbal descriptions “agree”, “slightly agree”, “slightly not agree”, “do not agree”. Based on the critical reading of the literature (Thomson et al., 2014; Ingham et al., 2015) and the expert panel meetings, we hypothesized that the instrument would have a four-factor structure with factors “conviction”, “personal benefit”, “personal resources”, and “time management” with lower scores meaning more positive outcomes on the respective scales.

Further instruments

We used the German version of the Maslach Burnout Inventory (MBI) to assess occupational burnout. The MBI comprises of 22 items to be scored on a 7-point-scale from “0-never” to “7-every day”. It consists of three subscales, namely emotional exhaustion (nine items) which measures exhaustion at work, depersonalization (five items), which measures loss of empathy and emotional distance to others, and personal accomplishment (eight items) which measures competence and positive attitude towards work. The three-factor structure was confirmed (Neubach & Schmidt, 2000). Cronbach- α of the emotional exhaustion scale was .85, of the personal accomplishment subscale .71, and of the depersonalization subscale just .48. Other studies found higher internal consistencies for this subscale with Cronbach-alphas of .69 and .86, respectively (Schwarzer, Schmitz & Tang, 2000; Gumz et al., 2013). Convergent and discriminant validity of the MBI could be demonstrated.

The Work Satisfaction Questionnaire is comprised of 17 items to be scored on a 7-point-scale from “1-very dissatisfied” to “7-very satisfied”. It has a five-factor structure with factors patient care (4 items, Cronbach- α = .76), burden (4 items, α = .79), income-prestige (3 items, α = .83), personal rewards (3 items, α = .71), professional relations (2 items, α = .66). Furthermore, a global item asks for the satisfaction with the current job situation. This item correlates with the subscale scores form .39–.71 (Bovier & Perneger, 2003).

Statistical analyses

There were up to three missing values on single items of the MoME-Q. They were replaced by the k nearest neighbor algorithm (kNN) (Beretta & Santaniello, 2016) using the R package VIM (Kowarik & Templ, 2016).

We conducted a confirmatory factor analysis with the R package lavaan (Rosseel, 2012). We used the robust Unweighted Least Squares Estimator (ULSMV), as this estimation method makes no distributional assumptions (Rosseel, 2012; Lei & Wu, 2015). Different model-fit statistics were calculated. The χ²/df ratio is a badness-of-fit-index as smaller values indicate a better fit (West, Taylor & Wu, 2015). Values around 2 signal a good model fit. The Root Mean Square Error of Approximation (RMSEA) is a population-based index that relies on the noncentral χ² distribution. It can be regarded as an “error of approximation” index because it assesses the extent to which a model fits reasonably well in the population (Brown, 2015). Values ≤ .08 are considered to indicate an adequate model fit (Browne & Cudeck, 1993). The standardized root mean square residual (SRMR) was calculated that measures the mean absolute value of covariance residuals (Little & Kline, 2016). Values below .10 indicate a good model-fit (Weiber & Mühlhaus, 2014). The Comparative Fit Index (CFI) and the Tucker Lewis Index (TLI) were not considered as they are sensitive to smaller sample sizes like ours in ULS estimation (Lei & Wu, 2015). The resulting items and scales were examined by parameters based on classical test theory like Cronbach- α, discriminatory power, average intercorrelations. Omega coefficients for the applied scales were also computed using R packages psych and GPArotation as they have known advantages over Cronbach’s- α (Raykov, 2001).

We used Hotelling’s T²-test from the R library “Hotelling” to compare different demographic groups on the scales of the MoME-Q (Hair, 2010). After a significant multivariate result univariate Welch t-tests were calculated to explore the analyses further. The univariate effect size partial Cohen’s d was then calculated with values of .20 representing a small effect, .50 showing a medium effect, and .80 a large effect (Grissom & Kim, 2012).

We used the Spearman correlation coefficient to calculate associations between the MoME-Q subscales and other instruments as most of the variables deviated significantly from the normal distribution (Kim, Kim & Ergün, 2015). Due to multiple testing, the significance value was adjusted by the Bonferroni correction (Bortz & Schuster, 2010).

Characterization of the study sample

The total population consists of 158 GPs. Of them, 85 (53.8%) took part in the study and completed the questionnaire. There are 64 male GPs (75.3%) in our sample. The gender distribution conforms to the proportions in the population in this specific area. Mean age of the participants is 53.5 years (SD 8.93) with a median of 54 years, a minimum age of 32 and a maximum age of 73 years. The majority (91.8%) are practice owners, work full-time (90.6%) and work in a group practice (67.1%). The average study participant works in private practice for 18.41 years (mean, SD 9.8 yrs.) with a range between 2 and 43 years. Most of them are specialized in General Practice (51.8%), whereas 24.7% are specialized in Internal Medicine and 20.0% have both specializations. The minority (3.5%) are “Praktischer Arzt” without any further specialization. This denomination has been disestablished and taken out the regulation for further education in 1992. It is notable that despite of all political obstacles 94% would become GPs again.

Looking at teaching, only 14% have an affiliation with a university. The majority (59%) has at least some teaching experience, for instance most GPs have seen students for a clinical elective (57%), whereas only 17% have seen students within a university primary care rotation program (Blockpraktikum). A minority of 11% has seen students for parts of the practical year and only 3% have ever given lectures or seminars at a medical school.

Table 1 summarizes the demographic characteristics of the study participants.

Motivation for Medical Education Questionnaire (MoME-Q)

The original version of the Motivation for Medical Education Questionnaire consisted of 28 items Table S1). After inspection of the statistical characteristics of the items and supported by the explorative nature of the study, we decided to exclude 4 of them from further analyses (Table 2). Further details are explained below.

The item “I hope to attract more patients being an “Academic Teaching Practice”” was initially considered to have a positive connation. The study results show, that GPs in the region of Siegen-Wittgenstein, threatened by a critical shortage, do not desire to treat more patients. Consequently, 76.5% of responses are in categories “3” and “4” which means that physicians do not wish to attract more patients for their practices as their current workload obviously is already high. The item is highly left skewed (p < .001).

The item “Students can support and relieve me in daily routine patient care” can be interpreted differently. On the one hand, the GP as a teacher should focus on teaching and support a good learning environment and not make use of the student’s work force in the first instance. On the other hand, integrating students in real patient care according to their state of knowledge can make sense and foster personal development. Furthermore, teaching students next to patient care in daily practice is also time-consuming. As the content of the item remained unclear, we decided to exclude it. The item “I hope that General Practice gets more attention if more GPs take part in medical education of students” is too general. Therefore, 76.5% of responses are in categories “1” and “2” which means that physicians hope that their specialty will get more recognition by participating in the education of students.

The item “I have made bad experiences with medical students in my practice in the past” has a mean of 3.81, a median of 4, and a SD of 0.15; 100% of responses are in categories “3” and “4” which means that physicians hardly ever made bad experiences with students. Most of the study participants did not make experiences with medical students at all, which made it impossible for the majority of study participants to answer this question.

Descriptive statistics of the remaining 24 items are displayed in Table 3. The numbering of the items in the article corresponds to the initial version of the questionnaire. As can be seen, a substantial number of items shows significant deviations from normality regarding skewness and kurtosis.

The remaining items were hypothesized to load on the 4 different factors conviction, personal benefit, personal resources, and time management. The confirmatory factor analysis with the robust ULSMV estimation method showed a good model fit: χ²∕df = 1.36, RMSEA = .066, SRMR = .096. Factors conviction and personal resources correlated .97 and there was also a high correlation between factors conviction and time management (r =  − .86). Model parsimony is a main target in confirmatory factor analysis. Highly correlating factors do not convey additional information. Therefore, factors conviction, personal resources, and time management were united into one factor called “commitment” and a new analysis postulating a two-factor model (“commitment” and “personal benefits”) was performed. The confirmatory factor analysis with the robust ULSMV estimation method again showed a good model fit: χ²∕df = 1.38, RMSEA = .067, SRMR = .098.

As shown in Table 4, all factor loadings are in the satisfactory range and the correlation between the two factors is also acceptable (r = .503). Therefore, this two-factor solution can be accepted and we calculated two subtest scores in the MoME-Q for further analyses. We have to mention that this is not a confirmatory but a model generating approach (Jöreskog, 1993) which means that this structure has to be confirmed in a new sample. This was done as the instrument was newly developed and there were only preliminary hypotheses about the structure of the questionnaire. After reversing items 18, 19, 20, 21, 22, 23, 24, 26, 27, and 28 the scale score of “commitment” was calculated.

The mean of the scale “commitment” was 31.0 (SD 8.4) with a median of 30, a minimum of 19, and a maximum of 53 (Fig. 1). Its distribution deviated significantly from a normal distribution: Shapiro–Wilk-Test, p < .001; Skewness, p = .003; Kurtosis, p = .35. Cronbach- α coefficient was .90, omega coefficient was .91, the average inter-item-correlation was .37. Discriminatory power of the items ranged from .31–.79. All values can be classified as satisfactory to high.

The mean of the scale personal benefit was 17.9 (SD 4.4) with a median of 18, a minimum of 10, and a maximum of 31 (Fig. 2). Its distribution mainly deviated significantly from a normal distribution: Shapiro–Wilk-Test, p = .02; Skewness, p = .02; Kurtosis, p = .19. Cronbach-α coefficient was .81, omega coefficient was also .81, the average inter-item-correlation was .34. Discriminatory power of the items ranged from .33–.65. All values can be classified as satisfactory to reasonably high.

Association with demographic characteristics

We median dichotomized age of physicians and then compared the two groups on the scales of the MoME-Q. The descriptive values are displayed in Table 5.

There was no significant multivariate effect (T²(2, 81) = 4.15, p = .14). The two age groups did not differ significantly on the scales of the MoME-Q.

We compared those physicians who had an authorization for performing practical education for future GPs with those who did not have this authorization on the scales of the MoME-Q. The descriptive values are displayed in Table 6.

There was a significant multivariate effect (T²(2, 82) = 7.32, p = .03). Univariate analyses revealed a significant difference between the two groups on the subscale personal benefit t(75.6) =  − 2.62, p = .01. A medium effect was shown by effect size Cohen’s d with .57. Those with authorization to perform practical education for future GPs hope to have more personal benefits than those who do not possess this authorization. No significant difference occurred on the scale commitment t(78.2) =  − 0.60, p = .55, d = .13.

We further compared those practices with a specialized practice nurse or with a practice nurse still in training with those who do not have a specialized practice nurse and who do not intend to have one in the future on the scales of the MoME-Q. This was done to investigate if there is a trend to support education of staff members in general. The descriptive values are displayed in Table 7.

There was a significant multivariate effect (T²(2, 81) = 7.93, p = .02). Univariate analyses revealed a significant difference between the two groups on the subscale commitment: t(81.9) = 2.82, p = .006. A medium effect was shown by effect size d = .62. Those physicians with a practice nurse show a higher commitment than those physicians who do not have a practice nurse and who do not have the intention to have a practice nurse in the future. There was no significant difference between the two groups on the scale personal benefit: t(75.8) = 0.85, p = .40 with a small effect size of d = .19.

We compared those physicians with experience in medical education with those who did not have experience in medical education on the scales of the MoME-Q. The descriptive values are displayed in Table 8.

There was a significant multivariate effect (T²(2, 82) = 25.93, p < .0001. Univariate analyses revealed a significant difference between the two groups on the subscale commitment: t(59.2) =  − 4.74, p < .0001. A large effect was shown by effect size d = 1.05. Those physicians with experience in medical education expressed a significantly higher commitment for medical education than those physicians without experience in medical education. There was also a significant difference between the two groups on the scale personal benefit: t(57.8) =  − 2.50, p = .015 with a medium effect size of d = .55. Those physicians with experience in medical education expect a higher personal benefit from medical education than those physicians who do not have experience in medical education.

Associations with burnout and work satisfaction

We calculated Spearman correlation coefficients between MoME-Q scales and subscales of the Maslach Burnout Inventory (MBI) (Table 9).

The MoME-Q subscale “commitment” correlated significantly with all three subscales of the MBI. Correlations were in the medium range around |.30|. Those physicians with higher commitment scores and therefore lower commitment to teach had higher scores on emotional exhaustion and depersonalization and vice versa. Those physicians with lower commitment scores and therefore higher commitment to teach had higher scores on the MBI subscale personal accomplishment and vice versa. The correlations of the MoME-Q subscale “personal benefits” with MBI subscales were around zero.

None of the correlations between the MoME-Q subscales and the subscales of the Work Satisfaction Questionnaire reached significance after Bonferroni correction (Table 10). The only tendency which can be reported is that those physicians with lower commitment scores and therefore higher commitment to teach are more satisfied with their current job situation and vice versa. A German version of the MoME-Q can be found in the supplement (Table S2).